本项目深入研究系统级故障诊断的基础理论与算法。主要内容包括：基于大型网的新系统诊断机制研究，遗传算法、模拟退火算法在系统诊断中的应用，诊断质量参数研究与应用，最小顶点覆盖、组合优化等一类NP难问题的研究，网络诊断仿真、实验研究等。随着网络和多机系统的不断发展，作为提高系统可靠性重要手段的系统级故障诊断研究将具有重大意义。

The principium of System-level fault diagnosis (SLFD) is to confirm status of nodes by analysis test symptom which is brought out by test among nodes in system. SLFD is one of key technologies for dependability of network and multi-computer system. After comparing and analysis the model and algorithm of existing SLFD, the theory of node grouping diagnosis is proposed including equation model, diagnosis performance metrics and diagnosis algorithm based on node grouping, which is suitable to large scale of network. The simulation system of SLFD has been developed which is used for estimating performance of different diagnosis algorithms through simulating the process of diagnosis. How to perform test and measurement on operating network system is also researched and how to guarantee quality of service of applications against nodes fault is discussed in this project. The group diagnosis is used in the network test and measurement system successfully. Some useful results on QKSORT algorithm, Boolean procedure, RTL circuit test and dynamic electrical current test are also presented.